Yesterday, a friend of mine linked me an article on the Civil War over at Time.com. It was good–really good. And I have to admit I was surprised. For whatever reason, I assumed that Time was a wishy-washy magazine full of puff pieces. So I snooped around the site a bit more, hoping to find more journalistic gold. And I definitely found some interesting articles. And while I have to admit they were worth reading, I also found some mushy thinking and scrambled reporting. I feel like the problems I ran across are worth discussing. (Hopefully you’ll agree, if not, sorry for the boring post!)
First up, I ran across an article on futurism, or, more specifically, on what some futurists call “the Singularity” (which is a bit confusing as this term has a number of other meanings). Essentially, this entrepreneur/computer engineer/futurist by the name of Raymond Kurzweil, among others, believes that in the coming decades, computers will achieve, and then surpass, human-level intelligence and come to dominate and/or fuse with us. He bases this belief on the well-known Moore’s Law, which points out the long term trend that “transistors that can be placed inexpensively on an integrated circuit doubles approximately every two years.” He has also developed his own interpretations of various trends, essentially showing that the processing power of computers has been increasingly at a similarly exponential rate.
And that trend is undeniable. We can look back at the processing speeds of computers over the last 50 years, and Moore’s Law holds up very well. And that progress has continued up to the present day. However, here’s where I think Kurzweil and others start to make problematic claims. I want to state clearly that I’m not saying that he, and other Singularitarians (the name they use to describe themselves) are wrong. Who knows whether their predictions will pan out? No one–and that’s the point. Assuming that, because a trend has persisted over some past period, it must continue along the same trajectory on into the future, forever, is fallacious.
Again, that doesn’t mean that the trend won’t continue, it just means we can’t assume that it necessarily will. This is easily demonstrable with all sorts of trends. For example, dinosaurs dominated the earth for about 150 million years. That’s a really, really, really long time. They had evolved some pretty impressive and unique traits that led to their dominance–their hips joined to their bodies in a different way than other reptiles, making them faster and more maneuverable. Many of them had the largest brain-to-body-size ratio of any animal at the time. Some may even have been warm-blooded, allowing them to out-compete other species. And in the late Cretaceous, dinosaurs were only honing these advantages. Any observer 65 million years ago would have assumed that this dominance would have continued for millions of years into the future.
Of course, we know that it didn’t. An asteroid strike (probably) struck the Earth at that time, and in short order, all dinosaur species were extinct. This is pretty common knowledge now, but it’s worthwhile to stop and think about this for a second. So there was a huge asteroid strike, ok–why should this result in the complete extinction of all dinosaurs? We might assume that top predators and the huge herbivores would have died out. They needed vast amounts of food every day, and in a darkened, volcanic world, it would have been difficult for them to survive? But why should the small, possibly omnivorous dinosaurs have died out? They were, arguably, some of the most “advanced” organisms of their time. I think most intelligent observers would have predicted their continued survival.
Now, I know very little about paleontology and dinosaur biology. So I’m certainly not going to hazard any guesses as to the ecological/metabolic/biological reasons for why dinosaurs indeed died out (and of course, some of them may actually have survived and evolved into dinosaurs, though the exact timing and causality of bird evolution is still a matter of great debate). But the point is simply this: there was a long and well-developed trend of dinosaur dominance, adaptation, and refining that ended abruptly. If we graphed the number of individual dinosaurs, dinosaur species, and the centrality of dinosaurs in the ecosystem, beginning with the first dinosaurs in the Jurassic period, we would see a steeply climbing line, which abruptly ended 65 million years ago. Trends do not have to continue. And for those who object by saying that the asteroid strike was a rare event, a sort of fluke–well, unlikely events occur. Their rarity, from a human perspective, isn’t terribly significant. When we make predictions about the entire future of the universe, our theories better be robust enough to deal with unlikely events–especially considering that we know full well that major catastrophes aren’t really unlikely, on a universal scale of reference.
Of course, you might argue that the dinosaur trend is so different in kind from the computer processor trend as to have no bearing on this debate. So let’s talk about something a bit closer to home: oil. If you graphed the amount of oil pumped out of the Earth, beginning in the middle of the 19th century until today, you would see an incredibly steep line heading up. At times the growth was certainly exponential. But not even the most optimistic oil investor thinks that line can continue forever. There is a finite amount of oil in the ground. We can debate how finite the supply is, but we really can’t debate whether it’s finite. The trend of a vastly expanding supply of oil absolutely will end, possibly in the next few decades. So as impressive as the exponential rise in the supply was, it would be completely foolish to think that expansion must necessarily continue indefinitely.
Now, computer processing power isn’t a finite resource lying in the ground. But computer circuits can only be so small. And their size is one of the most important factors in determining their speed. Computers can also be made faster by combining many processors on one circuit, so that they can share the processing load. Dual-core processors, quickly becoming standard today, are an example of this. But dual processing (or triple, quadruple, etc.) cannot yield infinite increases in speed either. Dual processing works best when there are a number of discrete tasks to do, so that some tasks can be given over to one processor and other to the second. But some processing tasks aren’t as conducive to this setup. Furthermore, there are necessarily inefficiencies in transmitting data between multiple processors. So while it seems likely we will continue to see huge increases in processor speed and power, the trend cannot possibly continue infinitely–there will be real limits.
Of course, just where those limits are is totally unknown. Perhaps we are very close to the limits and computers may not advance much beyond their capacities today; on the other hand, perhaps computers will continue to advance in speed and we will see supercomputers that are many times more intelligent than any human. We really don’t know, and as I said at the beginning of this piece, I’m certainly not saying that the futurist Singularitarians are wrong, but rather that their predictions are based on shoddy methodology. They’re manipulating statistics and ignoring fundamental tenets of the philosophy of science.
That said, they may end up being right anyway. And
many most all of them are certainly far, far more knowledgeable than I am about computer engineering and the other related fields salient to this discussion. But we’re not talking about matters they can have true expertise on, we’re talking about completely unknown possibilities many decades into the future. And I think that their approach is fundamentally unscientific–it seems to me their movement is carried forward more by a geekish love of technology (totally understandable) as well as, perhaps, a fear of death (also totally understandable). And I think the ideas they are advancing are interesting and very much worth talking about. I just wish the discussion could be carried with less adolescent enthusiasm and more respect for what science demands of our thoughts. I think all too often that futurists fail to keep any boundary between their desires and what could responsibly be called predictions. They are not the same.
I sort of wanted to end the post there, but another aspect of the whole discussion occurred to me while I was writing, and I think it’s worth adding: futurists/Singularitarians don’t seem to talk much about the environmental impacts/constraints of their vision for the future. Will there even be enough of electricity for massive computer programs in 50 years? If the prices of oil and coal sky-rocket, and societies have to choose between eating and moving around and funding energy-intensive projects like those the futurists envision, which do we think societies would choose? And which would the futurists advocate for?